Page 1 PART II QUANTIFICATIONAL LOGIC The language m of part I was built from sentence letters, symbols that stand in for sentences. The logical truth of a sentence or the logical validity of an argument, however, may hinge on the internal structure of the sentences; and it is the study of this structure that we now take up. Under the label quantificational logic, we consider logical systems that allow predication, i.e., the application of predicates to terms, and that also allow quantification, i.e., the application of quantifier phrases, such as some and all, to predicate expressions. Chapter 1 describes the syntax of a simple language of this kind and chapter 2 describes its semantics. Chapter 3 discusses various logical notions in the context of the new language. Chapter 4 presents an axiom system, PL, and illustrates its use in proofs and derivations. Chapter 5 contains a proof of PL's completeness, and chapter 6, an argument for its adequacy. Chapters 7 and 8 introduce systems of quantificational logic that can be viewed as extensions of PL. The proper treatment of some of these extensions is controversial, and we outline some of the more attractive alternatives. Chapter 1. Syntax We shall now set up a new language m(%) of predicate logic. Recall that m is the language of classical sentential logic and has an alphabet of sentence letters p1,p2,..., truth-functional connectives Z and ¬, and brackets ( and ). The alphabet of m(%) then consists of m ~ the connectives and brackets of plus the universal quantifier , the (object) variables v1,v2,..., n n and, for every non-negative integer n, the degree-n predicate letters P 1,P 2,... An atomic formula of m(%) is a degree-n predicate letter followed by n occurrences of object variables. The formulas of m(%) are determined by the following formation rules: (i) Each atomic formula is a formula. (ii) If A is a formula then so is ¬A. (iii) If A and B are formulas then so is (AZB). (iv) If A is a formula and x is a variable then ~xA is a formula. 0 For example P 6 is a an atomic formula (because it is a degree-0 predicate letter followed by zero 3 occurrences of variables) and P 1v1v2v1 is an atomic formula. Therefore, by (iii), 3 Z 0 ~ 3 Z 0 (P 1v1v2v1 P 6) is a formula and, by (iv), v2(P 1v1v2v1 P 6) is a formula. We have here posited a fixed and countable supply of sentence-letters and predicate- letters of arbitrary degree. We could generalize the approach as we did in chapter 8 and consider languages with different sets of sentence- and predicate-letters (and even of the variables). The correspondence between the new language and a natural language like English is not Page 2 as simple and direct as was the correspondence between m and English. Full discussion of this correspondence is postponed until the presentation of an informal semantics in the next chapter. But if F is a predicate letter and x and y are variables we can read Fxy as "F of x and y" and ~x as "for every object x" A formula is quantificational if it contains some occurrences of ~ and quantifier-free if it does not. A set of formulas is quantifier-free if all its members are. The formula A is a truth-functional compound of the formulas B1,...,Bn if A can be obtained by applying rules (ii)- (iii) to B1,...,Bn, i.e., if A is either one of the B1,...,Bn or the result of successively forming 3 Z ~ 0 disjunctions and negations of B1,...,Bn. For example ¬(P 1v1v2v1 v1P 6) is a truth functional 3 ~ 0 ~ 3 Z ~ 0 compound of P 1v1v2v1 and v1P 6 but v1¬(P 1v1v2v1 v1P 6) is a truth-functional compound of no formula but itself. The formula A is a universal formula if it is of the form ~xB for some variable x and formula B. Predicate letters of degree-one are sometimes called monadic. Notice that m(%) is a truth-functional language in the sense of chapter I.8. Its constituents are the atomic formulas and the universal formulas. All of the previous conventions concerning parentheses, abbreviations, naming and metalinguistic variables remain in force. In addition we shall use the boldface letters 'x', 'y' and 'z', with or without subscripts and primes, as variables whose range is the object variables of m(%), 'F', 'G' and 'H', with or without subscripts, as variables whose range is the predicate letters (of any degree) of m(%). Formula will now mean formula of m(%) unless otherwise stated. We add the following clauses to the definition of direct abbreviation: n n PLi) P > P 1 for all non-negative n, n n PLii) Q > P 2 for all non-negative integers n, n n PLiii) R > P 3 for all non-negative integers n, PLiv) v > v1, PLv) u > v2, PLvi) w > v3, PLvii) }x > ¬~x¬, n PLviii) Sx1...xn > S x1...xn 0 PLix) pj > P j for all positive integers j. The first seven clauses, unlike previous ones, involve direct abbreviation of non-formula expressions. They allow us to write a variety of formulas without subscripts on predicate letters and object variables. The symbol } in clause vii is called the existential quantifier. }x is read as for some x. Clause viii allows us to drop superscripts on the predicate letters in a formula. Doing so does not destroy uniqueness of unabbreviated form. The degree of the predicate letter can always be recovered by counting the variable occurrences that follow it. This abbreviatory device requires a modification on the definition of abbreviation. For under the previous 1 definition, Pv1v2 would abbreviate P v1v2, which would violate the property that all disabbreviation chains from formulas terminate in formulas. To prevent this, we modify the definition of abbreviation so that clause viii may only be applied when xn is not followed by a variable. A way to formulate the definition of direct abbreviation without modifying the definition of abbreviation is suggested in the exercises). The reader should keep in mind that when a formula like ~v(PvZ¬Pvv) is disabbreviated the two occurrences of P are replaced by Page 3 1 2 distinct predicate letters, P 1 and P 1. Predicate letters of degree zero are sentence letters. They constitute formulas themselves, without the addition of variables or other symbols. We use this terminology without prejudice to the question of whether there is a significant difference between sentences and predicates as ordinarily conceived. Notice that the sentence letters of m(%) are the 0 0 m capital letters P 1,P 2,..., whereas the sentence letters of were the lower case letters p1,p2,... But the final clause in the definition of direct abbreviation permits us to use the latter to abbreviate the former. This makes it possible to view m(%) as an extension of m as described in part I. Although the constituents of m are not constituents of m(%), they are abbreviations in m(%) of such constituents, and consequently every formula of m is a formula of m(%)-with- abbreviations. Some terminology regarding quantifiers and variables will be useful later. It is sometimes helpful to regard the expressions ~x and }x as unary connectives, like the negation sign. We refer to these expressions as (universal and existential) quantifier expressions (in x). Recall that the scope of an occurrence of a connective is the smallest occurrence of a subformula to contain the connective occurrence as a part. Thus the scope of an occurrence of a quantifier expression in a formula is comprised of the occurrence of the quantifier expression itself and a subformula occurrence, which may be called the proper scope of the quantifier expression occurrence. Every occurrence of x that is within the scope of an occurrence of a quantifier expression in x is bound by that occurrence. Every occurrence of a variable that is not bound by any occurrence of a quantifier expression is free. A formula is said to be closed if it contains no free occurrences of any variable, otherwise it is open. y is free at an occurrence of x in A if that occurrence does not lie within the scope of a quantifier expression in y (so that when the occurrence of x is replaced by a y, that occurrence of y is free). y is free for x in A if y is free at every free occurrence of x in A. For example, in the formula }u(~vPv Z Rvw), the first two occurrences of v are bound and the third is free. v is free for w, but u is not free for v. Two occurrences of x that are either both free or both bound by the same quantifier expression occurrence are linked. Two occurrences that of x that are not linked are said to be independent. For example in ~v(PvZQv) each of the three occurrences of v is linked to each other occurrence of v, whereas in ~vPvZ~vQv the first two occurrences of v are each independent of the last two occurrences. The variable occurrences can be partitioned into families so that any two occurrences in a family are linked and any two occurrences in different families are independent. In the previous example, the first two variable occurrences comprise one family and the third and fourth occurrences comprise a second family. Since the members of a family will either be all free or all bound by a single quantifier, the families themselves can appropriately be labeled free or bound.